On Mendel's Laws. 
The figures were deliberately chosen so as to make the value of 
C,, the same as in the last case, but the values of the remaining 
chances are totally different. The differences between successive 
chances run as follows in the two series— 
Failure of dominance. Failure of predetermination. 
•02500 -00500 
•00385 *00194 
•00056 -00075 
•00008 -00029 
In the first case the knowledge of the grand-parent makes a marked 
difference in the expectation as to the attributes of the offspring, 
but the higher ancestry are of very rapidly diminishing importance; 
in the second case the difference introduced by the knowledge of the 
grand-parent is small, but the higher ancestry are of greater relative 
importance. Moreover one could find a whole series of pairs of 
values for the proportions of /Ts contributed by ZTcells and 6-cells 
(instead of ’7 and *3), such that the chance of an A of unknown 
parentage producing an /4-form as offspring was 0*625, but for each 
of these pairs the remainder of the series of chances would be 
different. This is a greater generalisation than could be obtained by 
the former assumption of the failure of dominance alone. Given 
that the heterozygotes produce any fixed proportion of A’s, whether 
100, 80, 60 or 50 %, the whole of the series of ancestral chances 
is, in that case, determined, so that no two different ’series can be 
obtained starting from the same value. 
The preceding considerations should suggest to the “ Mendelian ” 
that it is a little futile to deny the fact of ancestral heredity when its 
existence is predicated by his own results, and to the biometrical school 
that they on their side should be rather cautious in drawing conclusions 
as to the processes that are or are not consistent with, and still more 
implied by, the existence of that phenomenon. I gave on pp. 205-6 supra, 
one theory accounting for the occurrence of ancestral heredity, viz.: 
the failure of “ predetermination,” and may confess that I had not 
then remarked that the application of Mendel’s Laws, without any 
modification, would lead to the same result; both theories implying— 
and this seems the one thing needful—the development of similar 
somatic characters from germ cells of different characters. One 
case of ancestral heredity arises directly from Mendel’s Laws, and 
a whole series of cases of a very general character indeed may 
be derived by supposing either dominance or predetermination of 
the somatic attributes to fail; the case of both failing would be more 
